Using in vivo EPID images to detect and quantify patient anatomy changes with gradient dose segmented analysis

Area of Science:

• Radiation Oncology
• Medical Physics
• Image Analysis

Background:

• Accurate dose delivery is crucial in radiotherapy.
• In vivo imaging offers real-time patient data.
• Predicting dose changes during treatment is essential for adaptive radiotherapy.

Purpose of the Study:

• To evaluate gradient dose segmented analysis (GDSA) combined with in vivo electronic portal imaging device (EPID) images for predicting planning target volume (PTV) mean dose changes.
• To retrospectively analyze clinical data for treatment site-specific deviations.
• To assess GDSA's utility in detecting day-to-day machine performance variations.

Main Methods:

• In vivo EPID transit images were analyzed using GDSA and gamma analysis.
• Phantom studies simulated errors (e.g., gas bubbles, weight loss, patient shifts).
• GDSA and gamma parameters were optimized to correlate with PTV mean dose changes from treatment planning system (TPS) data.
• Retrospective analysis of 852 patients over 23 months assessed clinical deviations and treatment site differences.
• Time-series analysis evaluated GDSA for tracking daily machine output.

Main Results:

• GDSA achieved a maximal R² of 0.90 in phantom studies for predicting PTV mean dose changes.
• For patient data, GDSA predicted PTV mean dose changes with 0.09 ± 0.98% accuracy and improved precision over gamma analysis.
• Over 95% of fractions showed deviations ≤2%; larger deviations occurred more in later fractions for head-and-neck and lung treatments.
• Daily averaged GDSA metrics identified machine output changes as small as 1%.

Conclusions:

• GDSA of in vivo EPID images is valuable for monitoring patient changes (e.g., weight loss, tumor shrinkage) during radiotherapy.
• GDSA provides a quantitative metric to flag clinically significant PTV mean dose deviations.
• Daily GDSA analysis can detect systematic treatment deviations due to machine performance issues.